To determine the OPN/αvβ3-mediated signaling mechanisms involved in prostate cancer cell migration, we generated different PC3 cell lines as described in the Methods section. OPN expression levels were detected by immunoblotting with an antibody to OPN (Figure 1A; top panel. Approximately 10–15 clones were analyzed for high expression of OPN by Western blotting with an antibody to OPN. Among the clones tested, we chose one that exhibited high expression (>10 times) of full length OPN (clone C1; lane 4) and mutant (RGDΔRGA) OPN (Clone C1; lane 5). Untransfected and pCEP-4 vector transfected PC3 cells (lanes 1 and 2) were used as controls. To reduce endogenous OPN in PC3 cells, we generated four different OPN SiRNA constructs in the pSilencer vector (lanes 6–9). Scrambled RNA constructs (lane 10) and vector (pSilencer 4.1-CMV neo vector) -transfected cells (lane 3) were used as controls. A significant decrease in OPN expression was observed in PC3 cells transfected with a SiRNA construct. This construct was targeted to human OPN cDNA sequence 383–403 (lane 7). Immunoblot that is shown in panel A was stripped and blotted with an antibody to GAPDH (Figure 1A, bottom panel) as a loading control.

Several clones (10–15) were isolated from the SiRNA construct targeted to human OPN cDNA sequence 383–403. OPN levels were significantly reduced in five SiRNA clones (C1–C5) as compared with scrambled RNA constructs and vector (pSilencer 4.1-CMV neo vector)-transfected cells. To determine the effects of OPN levels on MMP-9 activity, conditioned media was collected from four independent clones that express high level of full length OPN (B, lanes 3–6), mutant OPN (B; lanes 7–10) and five independent SiRNA clones (C; lanes 3–7). These conditioned media were used for gelatin zymogram analysis (Figure 1B and 1C).

We assessed the surface levels of CD44 in PC3 cell lines. PC3 cell lines were surface labeled with NHS-Biotin and equal amount of protein lysates were immunoprecipitated with an antibody to variant CD44 (vCD44 v3–v10) and actin. Actin antibody was used as an internal control for immunoprecipitation. Immunoblotting was performed with an antibody to streptavidin-HRP to visualize the surface expression of CD44 (Figure 2A). Surface expression of CD44 splice variants (vCD44) was markedly increased in PC3/OPN cells (Figure 2A, lane 2) as compared with PC3 cells (lane 1). Lanes 5 and 6 represent shorter exposure blots of lanes 1 and 2, respectively. Both PC3/OPN (RGA) and PC3/SiRNA cells exhibited a decrease in the surface expression of vCD44 (lanes 3 and 4). The blot shown in the Figure 2B was stripped and immunoblotted with an antibody to actin. Immunoblotting with an antibody to actin did not demonstrate a significant change in the levels (Figure 2B) despite changes in the surface levels of CD44 (Figure 2A). This indicates that equal amount of lysate proteins were used for immunoprecipitation. Our results demonstrate a significant increase in the surface expression of CD44 in PC3/OPN cells.

Since there was an increase in MMP-9 activity in the conditioned media of PC3/OPN cells (Figure 1B), we determine the surface levels of MMP-9 in PC3 cell lines (Figure 2C). Equal amounts of surface labeled lysates were immunoprecipitated with an antibody to MMP-9 and blotted with streptavidin-HRP. There are no changes in the surface levels of MMP-9 in PC3 cell lines, and only a single band of MMP-9 protein (85–90 kDa) was detected in all cells (Figure 2C). No changes in the surface levels of MMP-9 also serve as a biotinylation control for surface labeling. This indicates that the biotinylation reaction was equally efficient in all PC3 cell lines. Consistent with our previous observation 22, the observed increase in the surface level of CD44 is due to OPN/αvβ3 signaling as PC3/OPN (RGA) cells exhibited a decrease in CD44 surface expression.

Surface association of CD44 with MMP-9 was shown to provide a mechanism for tumor invasion in mouse mammary carcinoma and human melanoma cells 19. Expression of antisense CD44 and treatment of cells with anti-CD44 blocked hyaluronic acid (HA)-dependent MMP-2 secretion and, subsequently, invasion of a human lung carcinoma cell line 27. Given the above observations, which implicate CD44 and MMP-9 in tumor invasion, we sought to determine the surface interaction of these proteins in PC3 cell lines labeled with NHS-Biotin (Figure 3). In order to further corroborate the interaction of MMP-9 with CD44, PC3/OPN cells were also treated with a blocking antibody to CD44 (30 μg/ml; Figure 3, lane 5) prior to labeling with NHS-Biotin. Lysates were immunoprecipitated with a variant CD44 (vCD44) antibody (lanes 2–5) or a non-immune serum (NI, lane 6) and subsequently pulled down with streptavidin agarose. MMP-9 activity associated with the CD44-immunecomplex was analyzed by zymogram analysis as described in the Methods section. Zymogram analysis of one half of the immunoprecipitates is shown in Figure 3A. An increase in CD44-associated MMP- 9 activity was observed on the surface of PC3/OPN cells (Panel A, lane 4). Very minimal activity was observed in PC3 (lane 2) or PC3/OPN (RGA) cells (lane 3). We observed that pretreatment of PC3/OPN cells with a neutralizing antibody against CD44 clearly blocked the OPN-induced activation of MMP-9 (lane 5). PC3 cells secrete active MMP-9 (Figures 1 and 4) although the interaction of both pro- and active MMP-9 was observed on the cell surface. These results suggest that the surface expression of CD44 is required for the activation of MMP-9. Secretion of active MMP-9 by PC3 cells may offer a potential mechanistic explanation for the processes of ECM degradation and cell migration.

The other half of the immunoprecipitate was blotted with streptavidin HRP to detect surface levels of CD44 (Figure 3B). In agreement with the data shown in Figure 2, OPN over expression was associated with an up-regulation of vCD44 on the cell surface (Figure 3B, lane 1) as compared with PC3 cells (lane 2). This was not observed in either PC3/OPN (RGA) (lane 4) cells or in PC3/OPN cells treated with a blocking antibody to CD44 (lane 5). A significant decrease in the surface expression of CD44 was observed in these cells (lanes 4 and 5). The mechanism by which anti-CD44 blocks CD44 surface expression is not well understood. OPN/CD44 interaction was shown to be required for cell migration and CD44 synthesis 6. Based on this observation, we suggest that the failure of interaction of OPN with CD44 by anti-CD44 may reduce CD44 synthesis and surface expression. Changes in the levels of CD44 expression in PC3/OPN and PC3/OPN (RGA) provide insights into the role of the αvβ3 receptor in this process.

Stripping and reprobing of the above blot with an antibody to MMP-9 demonstrated the levels of MMP-9 interaction with CD44 (Figure 3C). MMP-9 levels corresponded to the surface expression level of CD44 in the indicated cell lines (panel C). Immunoprecipitation with a non-immune serum is shown in lane 3 (B and C). These observations suggest a biochemical pathway in which OPN mediated an increase in MMP-9 activity. This activity may take place via an increase in the surface expression of MMP-9 docking protein CD44 through αvβ3-mediated signaling 2228. These data demonstrate a role for CD44 in the surface localization of MMP-9.

Zymogram analysis of conditioned media of PC 3 cell lines treated with BPs: Giruado et al. have demonstrated that an amino-bisphosphonate zoledronic acid targets MMP-9 expressing macrophages and angiogenesis to impair cervical carcinogenesis 29. MMP-9 secretion was dose-dependently down-regulated by clodronate in human monocyte/macrophages 30. Bisphosphonate reduced the activities of MMP-2 and MMP-9 in PC3 cells 31. To this end, we investigated whether such inhibitory effects of MMP-9 activity by bisphosphonates would block cell migration and proliferation of PC3 cells. First, conditioned media from the indicated cell lines treated with or without BPs (50 μM each) such as alendronate (AL) and pamidronate (PA) for 48 h (Figure 4) were subjected to gelatin zymography. Consistent with the observation shown by others 31, treatment of the indicated PC3 cell lines with BPs, resulted in the reduction of MMP- 9 activity in all the cell lines. As shown in Figure 1, MMP- 9 activity was greater in PC3/OPN (Figure 4A, lane 5) than in PC 3 (lane 2) or PC3/OPN (RGA) cells (lane 8). The MMP-9 activity observed in alendronate-treated PC3 and PC3/OPN cells (lanes 3 and 6) is comparable to the activity observed in untreated PC3/OPN (RGA) (lane 8) or PC3/SiRNA (Figure 1C, lane 6) cells. It appears that the reduced cellular level of OPN itself has an inhibitory effect comparable to that of bisphosphonate. In the immunoblotting analysis with an antibody to MMP-9 (Figure 4B), no significant difference in the secreted level of MMP-9 protein was observed in the indicated PC3 cell lines treated with (lanes 4–6) or without (lanes 1–3) pamidronate for 48 h. MMP-9 activity but not the secreted level of MMP-9 is affected by BP treatment or OPN expression. The MMP-9 activity in the conditioned medium goes together with the surface expression levels of CD44 (Figure 2) in different PC3 cells. These data indicate that the actions of BPs or αvβ3-signaling involve MMP-9 activity as a principle therapeutic target for the control of cancer cell progression and metastasis.

Rho signaling pathway is a target for BPs. Rho signaling is essential for CD44 surface expression in osteoclasts 32. The fact that the expression of Rho GTPase induces clustering of CD44 and MMP-9 in the advancing lamellipodia of endothelial cells 33 prompted us to determine the effects BP on the localization of CD44 and MMP-9 in the indicated PC3 cell lines. The effect of BPs on the surface interaction of CD44 and MMP-9 was analyzed by immunostaining analysis with respective antibodies in cells that were neither fixed with paraformaldehyde nor permeablized with Triton X-100 (Figure 5). We found that PC3 cells express MMP-9 (green) throughout the cytoplasm; however colocalization (yellow) of CD44 (red) and MMP- 9 (green) was observed as clusters on the surface of both PC3 (Figure 5A) and PC3/OPN (Figure 5B) cells. Bisphosphonate treatment significantly reduces colocalization of CD44 and MMP-9 in PC3 as well as in PC3/OPN cells (A' and B'). As shown in Figure 2B, PC3/OPN (RGA) cells had a reduced surface level of CD44 (red). Bisphosphonate untreated (C) or treated (C') PC3/OPN (RGA) cells showed negligible colocalization of CD44 and MMP-9. The diffused distribution of MMP-9 was observed on the surface of these cells. The inhibition level of CD44/MMP-9 interaction in untreated PC3/OPN(RGA) cells (C) is relatively similar to the level observed in PC3 (A') and PC3/OPN cells (B') treated with BP. These results suggest that BP either partially or completely blocks the signaling pathway mediated by integrin αvβ3. MMP-s are Zn-dependent endopeptidases. In an attempt to further elucidate that the interaction of CD44 and MMP-9 on the cell surface is specific, we performed immunostaining analysis with an antibody to human Zip-1 (hZip-1). Human Zip-1 is a cell surface zinc transporter protein that was shown to be expressed ubiquitously on the surface of prostate cancer cells 34. Immunostaining analysis revealed that the interaction of hZip-1 with either CD44 or MMP-9 was negligible or not at all observed (data not shown). A significant increase in the coclustering of CD44 and MMP-9 on the cell surface of PC3/OPN cells as compared with PC3 cells (Figure 5A and 5B) suggest a plausible mechanism of activation of MMP-9 by CD44.

Analysis of morphology of live PC3 cells by phase contrast microscopy: Over-expression of OPN has been shown to augment the occurrence of multinucleated giant cells in pancreatic adenocarcinoma 35 and macrophages of rat glomerulonephritis 36. These observations prompted investigation on cell morphology in PC3 cell lines that have been used for the above-mentioned studies (Figure 6A). We demonstrated here that the number of multinucleated giant cells was increased in PC3/OPN cells. PC3/RGA cells have multinucleated giant cells at a level comparable to that of PC3 cells. PC3/SiRNA cells failed to display multinucleated giant cells (Figure 6A). In our observations, repeated passaging of PC3/OPN has no effect on the survival of multinucleated giant cells, indicating the genomic stability of this cell line.

Antibodies to the standard and variant (V3–V10) CD44 were purchased from Biosource International, Inc. (Camarillo, CA). Antibodies to MMP-2 and MMP-9 were purchased from Chemicon International, Inc. (Temecula, CA). Cy2- and Cy3-conjugated secondary antibodies were purchased from Jackson Immunoresearch Laboratories, Inc. (West Grove, PA). GAPDH antibody was purchased from Abcam, Inc. (Cambridge, MA). Biotin (EZ-link Sulfo-NHS-LC Biotin), streptavidin-HRP, and ECL- reagent were bought from Pierce (Rockford, IL).

Mutation in the integrin-binding motif (Arg-Gly-Asp [RGD] Δ Arg-Gly-Ala [RGA]) of human OPN 21 was generated using altered sites II in vitro mutagenesis system (Promega, Madison WI). Prostate cancer epithelial cells (PC3, CRL-1435; ATCC; Manassas, VA) were transfected with full-length (PC3/OPN), mutant OPN (PC3/OPN [RGA]) in pCEP4 vector, and vector without insert (pCEP4) with use of Lipofectamine 2000 (Invitrogen, USA) following the manufacturer's instructions.

The OPN SiRNA expression vector was generated using Seqwright DNA Technology Services (Houston, TX). A minimum of four SiRNA constructs for the OPN gene were generated to allow selection of one that expresses an optimal silencing effect. Scrambled RNA construct was used as a control for SiRNA constructs. For each construct, two oligonucleotides that encode sense and antisense sequences separated by a short spacer region (approximately 5–7 bp, forming the loop structure) were designed and synthesized. Location of the target sequence for the constructs are 126–146, 383–403, 649–669, 869–889 in OPN cDNA. PC3 cells were transfected with the SiRNA constructs and pSilencer 4.1-CMV neo vector (as vector control) using a silencer SiRNA transfection kit (Ambion, Austin, TX).

Individual clones (about 15–20) were isolated following transfection of PC3 cells with the above-mentioned OPN constructs. PC3 cells were transfected with use of Lipofectamine 2000 (Invitrogen, USA) following the manufacturer's instructions. After 48 h transfection, cells were maintained in medium that contained different concentrations of G418 sulfate, for selection. Cells were kept in 400 μg/ml G418 for 3 d followed by 200 μg/ml G418 sulfate for 7–10 days and continued in 100 μg/ml G418 for 20 days. After selection, G418-resistant clones were observed as cell clusters. Individual clones were isolated for each construct in 24-well tissue culture plates using cloning filters (Sigma). Once the small cultures grew to near confluency, they were transferred to 60 mm tissue culture dishes with RPMI medium containing 10% FBS. Clones were maintained individually for each construct. OPN expression levels were determined by immunoblotting analysis with a polyclonal antibody to OPN (generated in Sigma-GenoSys). Four clones (C1–C4) that express elevated levels of full-length and mutated OPN (RGDΔRGA) were selected. To reduce the endogenous levels of OPN, four different SiRNA constructs were transfected into PC3 cells. PC3 cells transfected with respective vectors and untransfected PC3 cells were used as controls. Maximum decrease was observed with a SiRNA construct targeted to sequence 383–403 in human OPN cDNA. Individual clones (C1–C5) that exhibited maximum reduction in endogenous OPN levels were generated from this cell line. PC3 clones that express the highest levels of OPN (full length and mutant) and a maximum reduction in the endogenous levels of OPN were used for all the experiments described here. These clones were designated as PC3/OPN, PC3/OPN (RGA) and PC3/SiRNA.

PC3 cell lines were cultured in Roswell Park Memorial Institute-1640 (RPMI-1640) media containing 10% FBS at 37°C (Gibco-BRL, Bethesda, MA). Bisphosphonates (BPs), such as alendronate and pamidronate were used for experiments. Stocks (1 mM) were made in PBS or sterile H₂O. Prostate cancer cells were treated with BPs to a final concentration of 50 μM for 48 h at 37°C. Some cultures were treated either with the blocking antibody to CD44 (30–50 μg/ml; BioSource AHS 4418) or GM6001 (15 μM at 37°C) for 48 h. Following various treatments, cells were washed three times with cold PBS and lysed with RIPA lysis buffer as described previously 22. Some cultures were used for immunostaining analysis as described below. For zymogram analysis, cells were kept in serum-free RPMI medium for 24–48 h in the presence and absence of various treatments. Conditioned media collected from various PC3 cell lines were subjected to gelatin zymography as described below and previously 10.

Surface localization of CD44 and MMP-9 was determined in cells that were not permeablized with paraformaldehyde or fixed with Triton X-100 by immunostaining with antibodies to variant CD44 (V3–V10) and MMP-9 as described previously 22.

After various treatments, cells were labeled with NHS- biotin according to the manufacturer's guidelines (Pierce, Rockford, IL). Briefly, cells were incubated with 0.5 mg/ml biotin for 30–40 min. at 4°C and washed three times with cold PBS. Cells were lysed with RIPA lysis buffer as described previously 22. Equal amounts of protein lysates were used for immunoprecipitation with an antibody to CD44 or MMP-9. Immunopecipitation of two proteins at the same time with two different antibodies (e.g. CD44 and actin as shown in Figure 2) was also performed. The immune complexes were adsorbed onto Protein-A sepharose beads or streptavidin agarose. Streptavidin agarose precipitates the biotinylated CD44 protein from the cell surface. The immune complexes adsorbed onto streptavidin agarose were washed three times with cold PBS and incubated with sample buffer with no reducing agent (β ME or DTT) for 10–15 min. at room temperature (RT). Zymogram analysis was performed as described below. MMP-9 activity associated with cell surface CD44 was analyzed by gelatin zymography as described below. The immune complexes adsorbed onto streptavidin agarose were also used for immunoblotting analysis with streptavidin-HRP to detect the cell surface levels of CD44. The immune complexes adsorbed to Protein-A sepharose beads were immunoblotted with streptavidin HRP to detect the surface level of CD44 or with a primary antibody of interest (e.g. actin). Immunoblotting with an actin antibody was used as a loading control. Prior to loading on SDS-PAGE, these samples were boiled with loading buffer containing βME. SDS-PAGE analysis and immunoblotting was performed as described previously 2223. Protein bands were visualized by chemiluminescence using the ECL-kit (Pierce, Rockford, IL).

Conditioned media collected from various PC3 cell lines were concentrated approximately 10-fold) with a centricon concentrator (Amicon, Beverly, MA). Ten micrograms of concentrated media protein was diluted to 20 μl in cold PBS. Immune complex made with CD44 (V3–V10) antibody, as described above, was also used for this analysis. Samples were mixed with SDS gel loading buffer with no reducing agent (βME or DTT) and incubated at RT for 10–15 min to dissociate immune complexes. SDS-PAGE containing 0.1% gelatin was used for electrophoresis. Samples were loaded without heating with sample buffer. After electrophoresis, gels were incubated overnight in a buffer containing 50 mM Tris-HCl, pH 7.6, 5 mM CaCl₂, 1 μM ZnCl₂, and 1% Triton X-100. Triton was used to remove SDS from the gel. Gels were then stained with coomassie brilliant blue for 2–3 h and destained with 7% acetic acid or water. Gelatinolytic activity was detected as clear bands in the background of blue staining 24.

Wound closure assay was performed as described previously 10. Cells were grown in 35 mm culture plates (Falcon) to near-confluent level in RPMI medium containing 10% fetal bovine serum (FBS; Cellgro). Multiple uniform streaks (~50 μm in width) were made on the monolayer culture with 10 μl pipette tips. The cells were immediately washed with RPMI-medium with 10% FBS to remove detached cells. Mitomycin (5 μg/ml; Sigma) was used in the medium to inhibit proliferation of cancer cells. Hence, the observed increase in PC3 cells is not due to increase in proliferation of cells. The experiment was performed with GM6001 (15 μM) and bisphosphonates (50 μM). The cell migration was monitored for 48 h, and pictures were taken at 0 h and 48 h time points with a digital SPOT camera attached to an inverted Nikon phase contrast microscope. Six to eight fields were analyzed, and the mean percentage of wound area covered by cells was calculated. The amount of wound closure remaining between the original wound widths was traced with SPOT Advanced 3.0.4 image software (Diagnostic Instruments, Inc.). The distance between the migratory cells in the wound area at a 90° angle to the wound margin generated at 0 h was measured using the same program. The distance of wound width was expressed in microns before and after migration. Following each experiment, cell viability was confirmed by trypan blue dye exclusion assay. Floating cells were marked with black arrows (Figures 2 and 4). Statistical analysis was performed as described below.

PC3 cells were plated in 6-well chambers at a confluence of 2.5 × 10⁴ cells per well in RPMI-1640 medium containing 10% FBS at 37°C for 18 hours. Cells were then treated with increasing doses of pamidronate as indicated. Cells treated with PBS were used as controls. Incubation was continued for 48 h. Cells were counted using a Neubauer's chamber at the end of 48 hours.

All values presented as mean ± SEM. A value of p < 0.05 was considered significant. Statistical significance was determined by analysis of variance (ANOVA) with the Bonferonni corrections (Instat for IBM; Graphpad software).